Phonology of English

HANDBOOK OF PHONOLOGICAL DATA
FROM A SAMPLE OF THE WORLD'S LANGUAGES

A Report of the Stanford Phonology Archive

Compiled and edited by

John H. Crothers
James P. Lorentz
Donald A. Sherman
Marilyn M. Vihman

1979 by The Board, of Trustees of the Leland Stanford Junior University

All rights reserved
Printed in the United States of America

PAGE 106 STANFORD PHONOLOGY ARCHIVE

VOLUME 1 — SEGMENT INVENTORIES, GENERAL COMMENTS > FOOTNOTES English

190

190 English

190 English

190 English

190 01 p-aspi rated 01 02 30 116
Ep] 03 60

Ep-unreleased] 61
Cp-nasal-releaseJ ot * 62
Ep-preglottalized) 05 63
(free )

190 03 b 01 02 03 08

Eb-unreleasedJ 61
Eb-nasal-release) 0I * 62
Eb-ha If -voice] 91
(free )
Ep-laxl 91
( free)

190 04 t-aspi rated 01 30 1,6

*Eglottal stop]
03 60

E t-unreleased] 61
Et-nasal-releasel 01 * 62
Et-dental) 61 *
( free)

Et-lateral-release) 07 65
It-preglottalized] 66
( free )
Et/sJ 1 * 8
( free)

[ t-retrof lex] 68

190 06 d 01 03 06 46

Ed-unreleased] 81
Ed-nasal-releasel 0l * 62
Ed-dentall
( free )

Ed-lateral-release] 07 65

Ed-retrof lex! 68

Ed-half-voicel 91

( free)

Et-laxJ 91

( free )

Ed/z] 1 * 8

(free)

190 08 k-aspi rated 01 30 1,6
jfc] 03 08 60

tk-unreleased] 61
Ek-nasal-release) Ql * 62
Ik-preglottalized3 69
(free)

190 09 g 01 03 08 08 ** 6
Eg-unreleased) 61
tg-nasal-release] 04 62
Eg-half -voice] 91>
( free )
Ik-laxl 91
( free)

190 10 t/s-hacek 01 09 10

tt/s-hacek-preglottalized) 70
( free)

190 11 d/z-hacek 01 09

t d/z-hacek-ha 1 f - vo i ce] 9 1
( free )

It/s-hacek-laxl 91
( free)

12 f 01

13 v 01

I v-approxi mantl 11 71
( free)

Ev-ha If -voice] 91
(free)
Ef-laxl 91
( free)

14 theta 01 12 13

15 eth 01 12 13
Eeth-approxlmant] 72
(free)

leth-half-voice] 91
(free)

ttheta-laxJ 91
(free)

16 s° 1 «

17 z 01 1*

Ez-ha If -voice] 91
( free)
Es-lax] 91
( free)

18 s-hacek 01 « 18

19 z-hacek ot 15 16
Es-hacek-lax3 91
(free)

20 x

( loan)

21 m 17 18 19 20

Em-labiodental) 73
Em-syllabic] 7t|
( free )

22 n 17 18 19 21

En-syllabic) 71 *
En-dental] 75
Cn-retrof lex) 68

23 eng 17 18 19 22
teng-syllabic) 711
( free)

24 I 23 24
Il-dental-velarizedJ 77 79
( free )

t 1-voiceless) 25 78
( free)

El-velarized] 79
El-velarized-syllabic] 79 80
Il-retroflex] 88

25 r-approximant-retrof lex 27
Ir-flap] 82

( free)

[r-approximant-retrof lex-voic
eless) 78
E r-f ricati ve-retrof lex] 82

26 h 18 29 35
Eh-voice] 83

Ec-f ricative) 8i *

27 glottal stop 67 90
( trans i t i onal > free )
*/t/aspi rated/

51 i-long
Eil 87

Eiota/iJ 1 * 0
( free )

52 iota

53 e-mid 38 *» *
tepsilon] 88
(allo>f ree)

54 ash 1 * 1

E ash-long] 87

55 schwa- long
*l schwal

56 schwa 37 87
(tag(+),allo)

^/schwa-long/
*[o-midJ
tepsi lon-dot) 85

57 ash-dot 1 * 2
*ta]

la-front) 88

58 u-long 32
lu) 87

tupsi lon/u] 43
(free)

59 upsi Ion

60 o-mid-long 31
Eo-midl 87 89
(alio, free)

*/schwa/

61 alpha 39

62 a-long 1 * 1 *
EaJ 87 88

*/ash-dot/

63 yod
tyod-voiceless) 78

64 w
*Ew-voicelessl

65 o-open 38

66 w-voiceless 1 *** 78
(tag(-),allo)

*/w/

PAGE t07 STANFORD PHONOLOGY ARCHIVE

VOLUME 1 — SEGMENT INVENTORIES, GENERAL COMMENTS , FOOTNOTES English

190

190 $a English $b British Received Pronunciation $d Germanic $e England $f 300 million speakers of

all English dialects $g Merritt Ruhlen $g Marilyn Vihman (review)

190 $a O'Connor, J.D. $b 1973$c Phonetics $g Middlesex: Penguin $q author is native speaker

190 $a Gimson, A.C. $b 1962 $c An Introduction to the Pronunciation of English $g London: Edward

Arnold

190 $a Halle, Morris $b 1975 $B original publication 1973, Linguistic Inquiry IV. 451-464 $c Stress

rules in English: a new version $e Essays on the Sound Pattern of English, eds. Goyvaerts and
Pullum, p. 261-276 $g Ghent: E. Story-Scient ia P.V.B.A.

190 $a Fudge, Erik Sb 1975 $c English word stress: an examination of some basic assumptions $e

Essays on the Sound Pattern of English, eds. Goyvaerts and Pullum, p. 277-323 $g Ghent* E.
Story-Scientia P.V.B.A.

190 $a Trnka, Bohumi 1 $b 1968 $B first edition 1935 Sc A Phonological Analysis of Present-day

Standard English $f Alabama Linguistic and Philological Series No. 17 $g Alabama: U. of Alabama
Press !

190 $a ACCEHT $A Fudge's account is essentially a drastic revision of Halle's system, which alters

many of the technical conventions, especially those connected with secondary stress, and
incorporates some clearly needed statements on the accentual properties of affixes. Fudge
replaces Halle's PSR and SSR by the simpler "general stress rule 11 (p. 320) by which the
antepenult (if there is one) is stressed when (a) the ultima is strong, (b) ultima and penult
are both weak; otherwise stress is penult. The numerous exceptions seem to be about the same as
for Halle's rules. Fudge divides suffixes into six classes, including "neutral" suffixes that
do not affect stress, stressed affixes, and four classes that require stress to precede by a
fixed number of syllables. Also the unstressable prefixes (cf. above on Halle's rules) are a
notable category,

190 $a ACCEHT $A The article by Halle represents one of the later stages in the effort by

generative phonologists, particularly Halle himself, to give rules for English stress. In
general the more traditional idea that stress is free is partly confirmed by this effort, since
the rules usually allow a couple of accentual possibilities for a word of any given segmental
structure. Due to the complexity of the rules and the variety of alternative suggestions, only
a brief summary is given here, concentrating on location of primary accent in the word.
Subsidiary degrees of stress, a major concern of Halle's, are ignored here. Halle distinguishes
between weak syllables (with a short vowel followed by at most one C) and strong syllables (all
others). The effect of the rules can be divided in two parts, depending on whether the vowel of
the word-final syllable is long or short. (Diphthongs count as long V; for the purposes of the
rule /i-long, u-long >o-long/ in open final syllables may count as short V.) (1) Short V in
final syllable: stress is antepenultimate if the penult is weak; the penult is stressed if it
is strong, and it is also stressed in di syllables. (p.265-the "primary stress rule n (PSR))
However, many verbs and some nouns stress a weak penult, and again verbs and some nouns stress
the ultima if it ends in a consonant cluster. Further exceptions are words stressing a weak
ultima. Also there are many words of both Germanic and non-Germanic origin taking stress on a
weak ultima, or weak penult, when an unstressable prefix immediately precedes. (2) Long V In
the final syllable: the "stressed syllable rule" (SSR) (p. 266) applies, putting stress on the
antepenult, or on the preceding syllable (if one exists) when the antepenult is weak. Many
words, particularly verbs, also nouns with certain final syllables (e.g. "-oon"), stress the
ultima instead.

190 $a ACCENT $A Gimson (p.216ff) finds that word accent is manifested primarily by pitch

prominence, also by stress, length, and vowel quality. As to the last category, /schwa/
generally occurs only unstressed, and long vowels and diphthongs are perceived as having a
certain degree of prominence, even without any of the other phonetic concomitants of accent. He
sets up three degrees of accent: (1) primary (tonic), carrying a pitch prominence, (2)
secondary, with qualitative or quantitative prominence, but having pitch prominence generally
only before primary accent, and (3) unaccented, (p. 222) $A In Gimson's inventory of accentual
patterns all words have one primary accent and at most one secondary accent. Most patterns of
over three syllables have a secondary accent. If there are two or more pretonic syllables, one
of them, two or more syllables before the primary, must have a secondary accent. Words with two
or more unstressed syllables between primary and a preceding secondary are compounds or have
primary accent on an affix .(e.g. "materialistic"). When three or more unstressed syllables
follow the primary, this is almost always due to unstressed affixes. Secondary accent on the
pretonic syllable is found only with words having primary accent on the second syllable.
Similarly secondary accent on the posttonic is only found word finally (except for certain
compound forms). Gimson regards word accent as fixed in the sense that a given word has one
accent pattern (or, in some cases, a choice of two patterns (p.227f)), and free in the sense
that accentual patterns exist with primary accent on almost any syllable of a word. Compounds
generally take primary accent on the first element and reduce the second accent to secondary,
but the reverse also occurs. A notable morphological stress pattern is alternation of many
disyllabic stems between final stress for a verbal form and initial stress for the
corresponding noun.

PAGE 108 STANFORD PHONOLOGY ARCHIVE

VOLUME 1 — SEGMENT INVENTORIES, GENERAL COMMENTS , FOOTNOTES English

190

190 $a INTONATION $A Within an utterance each accented word tends to constitute a rhythmic beat,

with beats occurring at "fairly equal intervals." Each accented syllable may have a pitch
contour, either falling or rising. Intonation patterns of sentences are related by Gimson to
the pitch contour of a single dominant ("nuclear") syllable. Falling intonation Is
"matter-of-fact, separative, and assertive." Rising intonation is "continuati ve, often with
overtones of politeness" etc. These patterns are found with both assertions and yes-no
questions, as well as other sentence types, (Gimson, p.234ff )

190 $a LOHG CONSONANTS $A Long C occur across word boundaries when identical C come together.

(Gimson, p. 151 )

190 $a HARGINAL SPEECH SOUNDS $A /x/ occurs in some Gaelic, German, and Welsh loans, but Is "of

very restricted occurrence." (p. 152) Extrasystemic sounds are the clicks It/s-clickl
("tut-tut") for disapproval and t t/l-f ricati ve-clickl ("gee-up") for urging on horses,
(p. 43-44) A trilled "r" is sometimes found "in highly stylized speech" for
/r-approximant-retrof lex/. (Gimson, p. 203)

190 $a PHONOLOGICAL UORD $A initial C* all but /eng, z-hacek/ $A initial CC: /s/ + tp, t, k) , /m,

n, 1, w/; /p-aspirated, b, k-aspirated, g, f/ + /I, r-approximant-retrof lex/*, /t-aspirated, d,
theta, eth, s-hacek/ + /r-approximant-retrof lex/; /t-aspirated, d, theta, k/ + /w/»
/p-aspirated, b, t-aspirated, d, k-aspirated, f, v, theta, s, m, n, 1, h/ + /yod/ $A Initial
CC* (in loans) /p-aspirated, t-aspirated/ + /s/; /s/ + /f/; /g/ + /z/; /b/ + /d/; /g, m/ + /w/;
/z/ + /yod/; /t-aspirated/ + /m/; /p-aspirated, k-aspirated, g/ + /n/ $A initial CCC: /s/ + tp,
t, kJ + /r-approximant-retroflex/5 /s/ + tp) + /l/; /s/ + tk) + /w/5 /s/ + tp, t, k, ml + /yod/
$A initial CCC: (in loans) /s/ + tk) + /!/; tp) + /s/ + /yod/ (Trnka, p. 43) $A final C: all but
/h/ $A final CC: (list restricted to stem-internal clusters) /s/ + tp, t, k) 5 tp, kl + /s/; tp,
k, f] + tt); /d/ + /z/; /m/ + tp), /f, z/; /n/ + It), /d, t/s-hacek, d/z-hacek, s, z, s-hacek/;
/eng/ + tkl ; /l/ + tp, t, kl , /b, d, t/s-hacek, d/z-hacek, f, v, theta, s> m/; /yod/ + It), /d,
s, z, n, 1/ $A final CC: (uncommon) /m/ + /b/5 /n/ + /theta, z-hacek/5 /l/ + /n/; /yod/ + /f/,
tk) $A final CCC: / m / + [p] + Et, s) ; /eng/ + tk) + tt, s) ; /l/ + tk) + tt, si; tk, 1, yod) +
/s/ + tt) ; /yod/ + /n/ + tt) $A final CCC: (uncommon) /n, 1/ + tt) + /s/ (Trnka*, p. 44) $A final
CCCC: Trnka does not find this within a stem, but it does occur by suffixation. See O'Connor,
p. 230.

190

$a SYLLABLE $A (C)(C )(C )V( V) (C ) (C ) (C )( C ) $A diphthongs: /iota, e-mid, upsilon, o-open/
/schwa/; /e-mid, ash-dot, o-open/ + /iota/; /schwa, alpha/ + /upsilon/

190 $a VOHELS $A The description of the vowels by Gimson goes into considerable phonetic detail,

and includes many dialect variants. Only the most prominent variants within RP have been coded
here.

190 01 $A "Fortis/lenis... corresponds auditorily to strong versus weak sounds, with the addition that
the lenis sounds are regularly shorter than the fortis ones. Acoustically the intensity of the
burst of the fortis plosi ves. . . is greater than that of tthe lenis plosi vesl . . .and the intensity
of the noise of fricative sounds... is greater for the fortis sounds than for the lenis."
(p. 130) The tense/lax opposition is manifested by a complex of factors, chiefly aspiration vs.
lack thereof before stressed vowels, voicelessness vs. voicing In intervocalic position with
the second vowel unstressed, and short vs. long preceding vowel (or sonorant) in syllable final
position. (Gimson, p.146ff)

190 02 $A "Occasionally the place of articulation is labio-dental when either /p-aspirated/ or /b/ is
immediately followed by /f/ or /v/." (p. 135)

190 03 $A "Sometimes... we find cases where the stop of the consonants is not completely formed, so
that air is never completely stopped from flowing and a very short fricative sound is heard
rather than a true plosive. This happens most often medially, between vowels." (p. 130)

190 0I » $A "The soft palate is lowered sharply so that the compressed air bursts out through the nose"
in the production of stop allophones with nasal release, (p. 134)

190 05 $A "The vowel of 'hope 1 is cut off by closure of the vocal chords; the tp] is then formed, the
glottal stop released and the tp) exploded normally. " (p. 136)

190 06 $A "/b, d, 3/ may be and often are voiced but they do not generally have full voicing. . .unless
they are surrounded by other voiced sounds." (p. 130-1)

190 07 $A "Instead of the tongue-tip lowering to release the compressed air, the sides of the tongue

lower and the air bursts over them, the tip remaining in position for the alveolar lateral til"
in the production of t t-lateral-release, d-lateral-release) . (p. 137)

190 08 $A Ik, g) are "made with the back of the tongue in contact with the soft palate, but the actual
centre of contact varies according to the surrounding sounds." (p. 137)

190 09 $A "The tongue-tip is against the back of the alveolar ridge, the front of the tongue rather

high in a secondary palatal articulation" in the production of /t/s-hacek, d/z-hacek/. (p. 138)

PAGE 109 STANFORD PHONOLOGY ARCHIVE

VOLUME 1 — SEGMENT INVENTORIES, GENERAL COMMENTS, FOOTNOTES English 190

190 10 $A /t/s-hacek/ "may have a little aspiration." (p. 138)

190 11 $A "In the case of the lenis /v/> there may quite often be no audible friction at all." (p. 140)

190 12 $A "The tongue-tip close to either the edge or the back of the upper incisors; the exact place
of articulation is not very important, and perfectly recognizable t theta) and [eth] sounds can
be made with the tongue-tip quite well back on the alveolar ridge..,. The sides of the tongue
are in close contact with the sides of the palate from back to front, ensuring that the
air-stream is funnelled to the point of maximum narrowing." (p. 141)

190 13 $A In the production of /theta, eth/ "the . . .friction. . A is) much less strong than for /s/ and
/z/." (p. 141 )

190 111 $A /s» z/ are "formed by the blade of the tongue making almost complete contact with the

alveolar ridge but leaving a narrow groove along its median line. This groove is considerably
narrower than the narrowing made for the non-sibilant /f, v, theta, eth/ sounds. As with
/theta/ and /eth/ the sides of the tongue are in close contact with the sides of the palate so
that breath is channelled through the groove and does not escape laterally. The teeth are
, always close together for /s/ and /z/ and it is noticeable that separating the teeth to any
extent immediately reduces the necessary sibi lance. . Differences in realization of /s/ and
/z/ are due to more or less grooving of the tongue-blade, leading to greater or less
sibilance." (p. 142)

190 15 $A In the production of /s-hacek, z-hacek/ "the tongue-blade is close to the back of the

alveolar ridge and there is grooving similar to that of /s/ and /z/ though not so narrow....
The front of the tongue is raised rather high in the secondary palatal articulation." (p. 143)

190 16 $A "Some speakers have a degree of lip-rounding for t/s-hacek, z-hacek/I *n all positions; for
others lip position is dictated by neighboring sounds." (p. 143)

190 17 $A "Mostly the allophones are fully voiced; after voiceless plosives the nasal consonants may
lose some voice, but never all or even most of it." (p. 145-6) +-;~^.<

190 18 $A "[Nasal consonants and /h/] do not participate in the fortis/lenis correlation." (p. 143, 145)

190 19 $A "tWith the nasals] occasionally, as with the plosives, the stop in the mouth may not be
quite completed, in which case a nasalized f Motionless continuant results." (p. 146)

190 20 $A "/m/ is less often syllabic t than /n/3 but may be so" word finally after plosives, (p. 146)

190 21 $A /n/ is produced with "the tongue-tip on or near the alveolar ridge," (p. 147)

190 22 $A "/eng/ is rarely syllabic." (p. 146)

190 23 $A "Most allophones of /l/ are voiced throughout, but if /!/ follows one of the voiceless "
plosives... part or all of the lateral articulation may have no voice." (p. 148)

190 24 $A "The sides of the tongue are not in contact with the sides of the palate along all their
length and air is therefore free to pass over the sides of the tongue" in the production of
/!/. (p. 148)

190 25 $A When /l/ is voiceless "the breath passing between the sides of the tongue and the palate may
and often does cause lateral friction; otherwise, /I/ allophones are non-fricative." (p,148).

190 26 $A "When /l/ is syllabic it is of much the same length as the vowels, when non-syllabic it is
shorter." (p. 149)

190 27 $A "The tongue-tip is fairly close to the back of the alveolar ridge, the sides of the tongue
are in contact with the sides of the palate and there is lateral bunching of the tongue, which
seems to be an important feature of /r-approxi mant-retrof lex/, since the correct effect can be
given without raising the tongue-tip, provided that the bunching is present." (p. 150)

190 29 $A "/h/ sounds. ..are simply strong voiceless versions of a following vowel and therefore do not
generally have a close enough tongue position to produce that kind of local frication. There
may be a little local friction between the vocal cords, but mostly what we hear is a more
generalized friction of breath passing through a relatively open vocal tract." (p. 143-4)

190 30 $A The aspirated stops occur "before a stressed vowel in the same word" except after /s/.
(p. 132)

190 31 $A /o-mid-long/ "often approaches" cardinal [o] , (Gimson, p. 109)

190 32 $A Gimson indicates considerable centralization for /upsilon/. (p. 112) Centralization of
/u-long/ is also found with "certain types of RP." (p. 114)

190 35 $A "There are as many allophones of /h/ as there are vowel sounds which can follow it." (p. 144)

PAGE 110 STANFORD PHONOLOGY ARCHIVE

VOLUME 1 — SEGMENT INVENTORIES, GENERAL COMMENTS, FOOTNOTES English 190

190 36 $A /o-open/ only occurs in the diphthongs /o-open. iota/ and /o-open, schwa/ (the latter being
quite rare in RP as it is increasingly replaced by /o-mid-long/ ) . (Gimson, p. 110)

190 37 $A /schwa/ occurs only unstressed. (Gimson, p. 87) Gimson (p. 119) gives the height as "between
half-open and half-close. 11

190 38 $A Gimson (p. 123) indicates that in the RP diphthong /e-mid.1ota/ the first element may range
from higher-mid to lower-mid, and may be centralized.

190 39 $A /alpha/ has "slight open lip-rounding. n (Gimson, p. 107)

190 1,0 $A /i-long/ is "most often realized as a slight diphthong in RP, though it may also have a
monophthongal realization." (p. 154, Gimson, p. 95)

190 1,1 $A "There may be a diphthongal glide... to a more central [position) 11 in the production of
/ash/, (p. 157, Gimson, p. 101) Also see Gimson (p. 99) on similar variation of /e-mid/.

190 ** z $A "Older speakers tend to favour the back part lof the central area)... and younger speakers
the front part" in producing /ash-dot/, (p. 158)

190 1,3 $A "In RP, /u-long/ most often has a diphthongal realization..., but it may be given a

monophthongal pronunciation slightly lower and more central than cardinal tu) ." (p. 162, Gimson,
p. 114)

190 ^ $A Both O'Connor (p. 159) and Gimson (p. 87) put /a-long/ slightly back of the central position.

190 ** 5 $A /w-voiceless/, a "fortis voiceless labio-velar fricative," is found Initially as a phoneme
in contrast with /w/ "amongst careful RP speakers and regularly in several regional types of
speech.... The use of /w-voi celess/ as a phoneme has declined rapidly." (Gimson, p. 212)

190 46 $A Gimson (p. 153) mentions that all plosives may have affricated realizations, with a very
brief fricative off-glide. He especially emphasizes this for the alveolars. (p. 158)

190 60 $A A stop has "little or no aspiration [when! it is followed by an unstressed vowel After

/s/ none of (the stops) has aspiration." (p. 132)

190 61 $A The stops are unreleased before another plosive consonant and "in final position in an
utterance." (p. 133-4)

190 62 $A "When any of the plosives is followed by any nasal consonant the explosion takes place
nasally." (p. 134)

190 63 $A "The glottal stop. ..often figures in the realization of /p-aspi rated/ (never of /b/) in RP
and other accents.... tltl occurs frequently before a pause.... The same sequence of events ,is
even more common in RP and many other accents before consonants." (p. 136)

190 6I * $A /t-aspirated, d/ "may be dental when a dental fricative follows." (p. 136)

190 65 $A /t-aspi rated, d/ have a lateral-release "when /l/ follows." (p. 136)

190 66 $A tt-preglottalized) as an allophone of /t-aspirated/ "occurs fairly commonly before pause and
more commonly before consonants." (p. 163)

190 67 $A "Before consonants it is not uncommon in RP to find (glottal stop) alone for /t-aspi rated/,
particularly if the following consonant is a stop." (p. 137)

190 68 5A /t-aspirated, d, n, 1/ are realized as retroflex "when /r-approximant-retrof lex/ follows,"
(p. 136)

190 69 $A "Glottalization of /k-aspirated/ is frequent in RP... before pause and before consonants."
(p. 138)

190 70 $A /t/s-hacek/ is "frequently glottal ized. . .before pause and before consonants..., but unlike
the plosives it is frequently glottalized before vowels." (p. 138)

190 71 $A /v/ is realized as t v-approximant) "quite of ten, . .particularly between vowels." (p. 140) (Not
mentioned by Gimson. )

190 72 $A "Like /v/, /eth/ is quite often realized intervocalically as a frictionless sound rather
than a fricative." (p. 141 ) (Not mentioned by Gimson.)

190 73 $A /m/ is labiodental before /f/ or /v/. (p. 146)

190 7 *» $A "/n/ is often syllabic. . .when no vowel intervenes between /t-aspirated/ or /d/ and (word

final! /n/. n Syllabification of /m, eng/ is much less common, but may also occur In comparable
environments, (p. 146)

PAGE 111 STANFORD PHONOLOGY ARCHIVE

VOLUME 1 — SEGMENT INVENTORIES, GENERAL COMMENTS, FOOTNOTES English 190

190 75 $A /n/ Is denial "when. followed by the dental fricatives /theta/ or /eth/. 11 (p. 147)

190 77 $A "Dental laterals may occur before /theta/ and /eth/>" (In which position they are also
velarized). (p. 148)

190 78 $A Li quids and glides may be voiceless when they follow "one of the voiceless plosives."
(p.148f> Gimson, p. 146)

190 79 $A /l/ is velarized when not preceding a vowel, (p. 148-9)

190 80 $A /l/ is syllabic when word final after an obstruent, (p. 149)

190 82 $A /r-approximant-retroflex/ may be realized as tr-flap] "in RP and other accents. . .1 when)
preceded by a short, stressed vowel and followed by an unstressed one, but this is not very
common." (p. 150) It is a fricative when preceded by /d/. (Gimson, p. 202)

190 83 $A /h/ is voiced "between voiced sounds." (p. 144)

190 8l » $A /h/ is realized as tc-f ricative) "when the following sound needs a high tongue position."
(p. 144)

190 85 $A /schwa/ is realized as t epsi lon-dot) in final position. This also applies to /schwa/ as the
final element of diphthongs. (Gimson, p. 119, 136)

190 8e $A /e-mid/ is realized as lepsilonJ as first element of the diphthong /e-ml d. schwa/. Before
/r-approximant-retroflex/ this diphthong frequently reduces to tepsilon). (p. 171)

190 87 $A The long vowels are shortened, before syllable-final voiceless consonants. (Gimson, p. 90) The
short vowels are also said to be longer before voiced than voiceless consonants (p. 100), but
Gimson's transcription and examples reflect this only for /ash/, which is longer than the other
short vowels. In the latter case the long phone is taken as variant of a basically short
phoneme. Gimson also treats all diphthongs as essentially parallel to the long vowelsas
regards phonetic length, with the first element shortened in syllables closed by voiceless
consonants. However the diphthongs are analyzed as clusters of two short vowel phonemes, not as
long vowel plus short vowel. Since length variations in the short vowels (other than /ash/) are
not coded here, those of the diphthongs are not either.

190 88 $A In the diphthongs /ash-dot. iota/ and /ash-dot. upsilon/ the RP norm for the first elements
are approximately [a-frontJ and tsl respectively. (Gimson, p. 124, 131)

190 89 $A Some RP speakers have to-mid] as first element of the diphthong /schwa, upsi Ion/. (Gimson,
p. 129)

190 90 $A (glottal stop) may be inserted between vowels at syllable boundary, and also word initially
before a vowel in emphatic speech. (Gimson, p. 163)

190 91 $A The voiced obstruents may be partly or wholly devoiced initially -in a stressed syllable and
finally. (Gimson, p. 148, 170, 174) The status of /z-hacek/ with respect to this rule is
marginal, since it does not occur initially, and is often replaced by /d/z-hacek/ finally,
(p. 185) However a voiceless variant is listed on p. 174.